Subject: Re: street performer protocol
From: Crispin Cowan <>
Date: Mon, 22 May 2000 00:42:01 +0000

"Stephen J. Turnbull" wrote:

> >>>>> "Crispin" == Crispin Cowan <> writes:
>     Crispin> "Stephen J. Turnbull" wrote:
>     >> The analogy is that Crispin is claiming that the Red Hat and
>     >> other Linux distribution successes are innovation (in software)
>     >> driven,
>     Crispin> Where did I say that?
> You denied that they were "software supermarkets" in a distribution
> industry and that their development efforts were crucial to their
> success.

I'm getting whip-sawed back and forth between two extremes:  major/successful
Linux distributors are neither purely innovation-based, nor pure commodity
distributors.  They are just about exactly in the middle:

   * They must be efficient at distributing the commodities. A lot of the
     value in a fresh Red Hat/SuSE/Turbo/Caldera/etc. CD is that it collects
     the latest stable version of many programs.  Related to this, the vendor
     should stay on top of vulnerabilities and bugs, and release erratta in a
     timely fashion.
   * They must also innovate:  much of the value in choosing one of these
     vendors over another is the unique software that they provide with their
     distribution.  Red Hat, Corel, and Turbo stand out here:
        o Red Hat:  developed RPM
        o Corel:  provides Word Perfect Office
        o Turbo:  TurboCluster, even if it is a separate product and not part
          of their standard distribution.

>     Crispin> If you read the history of Red Hat, you discover that it
>     Crispin> was almost entirely funded by revenue until late 1998 or
>     Crispin> early 1999.
> Huh?  They published their books?  Where can I get hold of those?

I dunnow about published books; I heard this informally.  But they didn't
have major investors until 98, and they did have major revenues.  Book
publishing should have happened as part of the IPO prospectus.

>     Crispin> Now reverse the situation: if capital for Linux startups
>     Crispin> dries up and blows away, who will survive?  The companies
>     Crispin> with the biggest revenue streams.  Relatively new me-to
>     Crispin> Linux startups that have neither technology nor revenue
>     Crispin> (because they invested the capital in marketing) will
>     Crispin> collapse.
> Then why did you bring up the "hundred Linux distribution startups
> that don't have to reinvent the wheel" if 99 of them are going to
> collapse?

Read the qualifiers above:  "those that have neither technology nor revenue"
will fail.  Niche specialist distros that do have a special technology to lay
atop the commodity open source substrate will survive selling their
high-value specialty thing with relatively low development costs compared to
fully proprietary competitors, who have to build a complete thingie top to

IMHO, this is what is driving the current rush towards "Embedded Linux":
Linux-based (or other open source) embedded system developers will be able to
massively under-cut proprietary OS vendors.

>     Crispin> How do you figure that?  Basic development is what makes
>     Crispin> a given Linux product better than the next one.  A Linux
>     Crispin> vendor can't sit still on technology, or they get
>     Crispin> bypassed and forgotten.  This is what happened to
>     Crispin> Slackware and *BSD :-)
> Both of which still exist in their niches, as far as I know.

It's hard to kill off a hobby :-)  Actually, this illustrates how specialty
FSBs can survive:  because they can build a valuable (read: high
functionality) product with relatively little development costs, they can
survive indefinitely on scant revenues.

>     Crispin> Just about every industry except corner stores
>     Crispin> consolidates, and even the corner stores are
>     Crispin> consolidating into francise brands like 7-11.  Of course
>     Crispin> the Linux market will consolidate.
> Again, this is the conventional wisdom; it doesn't explain how those
> 100 startups are supposed to prosper.

They're not:  as far as I can tell, many Linux distributors are not following
the above specialty business model, and think that they can get by on a
just-like-Red Hat business model.  I think that they are wrong, and they will
fail.  Those that go specialty will survive, at least for a while.

Naturally, the consolidation effect will move down-scale, following the niche
marketeers.  2000 will shake out the me-too Linux distributors, and late
2000/early 2001 will shake out the embeded Linux vendors.  Small vendors with
nothing will just fail, while small vendors with market share (revenue) or
technology will get acquired.

>  Or are they all academics with
> tenure on sabbatical, so they don't care?

Many of the teeny distros appear to be student projects, so they don't care.

>     Crispin> I completely disagree.  Open source system ventors will
>     Crispin> find it easy (verging on trivial) to crush closed-source
>     Crispin> competitors in these emerging markets, precisely because
>     Crispin> they do not have to invest $millions in re-inventing the
>     Crispin> wheel yet again.
> Ah, ever the optimist!  "Crush"??!?

Yes, crush.  Consider Lineo vs. QNX:  For the embedded/non-RT market, Lineo's
business model is far more attractive than QNX's, even though QNX's
technology is superior, because Lineo's cost base is lower for similar

The RT market is a little different.  Linux does not readily lend itself to
real-time, so a much greater investment in technology is required to produce
a viable Linux/RT product.  There are two or three players in that market,
and it will be interesting to see how they do against the likes of QNX and
VxWorks.  Lynux Works (nee Lynx Real Time Systems) is taking an interesting
position atop the fence.

> We have perhaps heard of "core competence"?  Even GE claims to have a
> core competence;

David Letterman? :-)

Crispin Cowan, CTO, WireX Communications, Inc.
Free Hardened Linux Distribution: